Bob Parkinson <rwp@biome.ac.uk> writes:
> Is there are simple rule for when the planner switches over from doing an
> index scan to a seq scan?
Sure: when it thinks the seqscan is cheaper ;-)
The cost models on which this is based are not so simple, but you
can study them if you like, see src/backend/optimizer/path/costsize.c
regards, tom lane